Given the two design options available, a greenfield deployment will be faced with a choice regarding the most appropriate vCloud NFV design to use. As an engineering decision, differences between the two designs and their potential use cases must first be understood. This section of the document provides considerations to guide CSPs in choosing between the two-pod and three-pod designs.
The vCloud NFV architecture ensures that key capabilities like secure multitenancy, integrated operational management, and carrier grade readiness are untouched by the choice of pod design. Neither choice will influence the ability of the NFV platform to support virtualized network services.
Two-pod design provides the VNF tenant with great flexibility in how to set up their virtual networking, both internally to the VNFs, and with external networks. Tenants can set up network and security services with minimal dependency on cloud networking administrators. This translates to a simplified VNF onboarding experience and great autonomy for the VNF tenant. To some CSPs, the ability to provide VNF vendors with a secure tenancy where the VNF can be configured and prepared to the tenant's liking is beneficial. Other CSPs will prefer to maintain control of all networking-related configurations, and may want to separate the VNF tenant from its external networking capabilities.
Another difference between two-pod and three-pod design is the footprint required for deployment. Two-pod design requires a smaller number of hosts, racks, and ToR switches than three-pod design. This means that distributed deployments such as those used in micro data centers or telecommunications central offices, where space and cooling is at a premium, may benefit from two-pod design. Some enterprise service use cases, such as premise-based virtual Customer Premise Equipment (CPE), also benefit from the smaller footprint of two-pod design. In these use cases, administrative boundaries can be made very clear by mapping the collapsed Edge / Resource pod to a single rack. In use cases where space is ample and virtual network functions perform a centralized role, maintaining the functional separation between the three pods is beneficial.
Capacity planning and scale up operations are natural and straight forward in three-pod design. With this design, each pod scales independent of the others. In two-pod design with both edge and resource functions sharing the same pod, as VNFs are added, careful consideration must be taken of the resources available to edge function operations. All the tools required for capacity planning and proactive resource usage monitoring are provided with vCloud NFV, and tools to migrate VNFs to necessary resources are also available.
As two-pod design facilitates greater autonomy to tenant operations, some advanced networking topologies will require the involvement of other functions and administrative domains. Depending on the type of communications to be used in the NFV environment, this aspect must also be taken into consideration.
